ALWAYS IMPERFECT – EPISODE 4 – STOP THE WOKE AI!

While the US focuses on putting the brakes on ‘woke AI’ (whatever that is) the rest of the world’s nations confront thorny questions about how to regulate a technology that’s both moving very quickly and lacks any clear definitions. Could regulators strangle AI in the process of regulating it? Or will innovators it outpace all efforts to contain it?  At the intersection of commerce and geopolitics, we speak with researcher Kate Carruthers, who puts these questions into a global context.

Meanwhile, Anthropic vs USA is also happening – can AI firms ever self-regulate?

ALWAYS IMPERFECT – Episode 3 – A BETTER WAY

Used well, artificial intelligence can automate labor-intensive and rote processes, freeing people for the work they want to do. That’s the theory, anyway – but what about the practice? SUPAHUMAN founder and CEO Dave Howden shares his experiences helping businesses adopt AI within their workflows – and admits, in order for that to work, he needs to be the ‘dumbest person in the room’. Humility and artificial intelligence – is that the foundation of ‘a better way’?

Full disclosure: my consultancy, Wisely AI, has a partnership with SUPAHUMAN.

ALWAYS IMPERFECT – Episode 2 – VIBE SLOGGING

Can a computer program a computer as well as a human being can? Artificial intelligence enabled a quantum leap in the quality of the tools programmer use to write code – but they’re delicate. Push them too hard and they break. Even when they work they can write reams of code that no human can make heads or tails of. John Allsopp joins us to investigate whether programmers will soon become obsolete – or whether they’ll kept around to clean up AI-generated messes. Is AI making the discipline of software engineering any better – or is that just a story we’re telling ourselves?

An attorney in California just got whacked with a USD $10,000 fine for submitting a briefing with hallucinations to a court.

Simon Willison’s most excellent blog is here.

ALWAYS IMPERFECT – Episode 1 – NO MAGIC WANDS

Artificial intelligence may be amazing but it’s always imperfect. Anyone trying to use AI professionally lands on the horns of a dilemma: will a productivity increase gained through automation represent any savings, after factoring in the extra supervision needed to use these amazing (but unreliable) new tools? In our first episode we chat with Drew Smith, co-founder of Wisely AI, a firm dedicated to helping businesses use AI safely and wisely. (I was the other co-founder!) What did we learn from clients trying to put AI to work – but only rarely finding the tools on offer fit for purpose?

Read the MIT report mentioned by Mark here.

Read The Register article mentioned by Drew here.

Two Futurists Go ‘Vibe Forecasting’

Whenever fellow futurist and longtime friend Rob Tercek and I get together, we go crazy deep on the future: What happens when lazy humans outsource their thought process to machines? You get a society that vibes its way into a blurry, sub-optimal future. The surge of slop means that AI is creating more work for humans instead of stealing our jobs. I join Rob on his podcast The Futurists to cut through the hype and the skepticism about modern tech, calling out AI mediocrity and reminding us why human learning remains painful. Topics include: ChatGPT psychosis, why we need a generation of PhDs to revisit eternal questions about Truth, the perpetual dawn of AGI, what happens to the political economy when the populace is siloed into bubbles, why Kremlin propagandists produce propaganda for machines instead of people, why experience cannot be generated, why autocracies need accurate data, and the real reason why people get lost in untruth. Just some light listening for you.

Big thanks and more at The Futurists podcast – https://thefuturists.com

Addiction is the business model

Mark joins RNZ Nine to Noon host Kathryn Ryan to delve into the sudden phenomenon of ‘viral’ AI videos – getting better at capturing your eyeballs. Did you see those cute bunnies bouncing on a trampoline? Never happened – but made you look. The fusion of TikTok and AI video looks to be potent – and addictive. Next up, shocking and sad story of Adam Raine, a 16 year old who confided his suicidal thoughts to ChatGPT. ChatGPT aided him in his eventual suicide – so Raine’s parents are suing OpenAI, and just after that lawsuit made news, OpenAI announced long-overdue parental controls on ChatGPT. Finally, we delve into… wait, why am I writing ‘delve’ so much? It’s because ChatGPT’s peculiar word choices – “delve”, “intricate” and so on – have invaded our speech!

Does ChatGPT Make Us Stupid?

A recent study from MIT shows students tasked with writing essays showed significantly lower levels of brain activity when allowed to use ChatGPT. Mark reviews the implications of this finding with RNZ NineToNoon host Kathryn Ryan, going on to reveal how an internet filled with AI-generated content becomes increasingly ‘toxic’ for those same AIs. Plus, would you agree to be ‘re-animated’ as an ‘AI ghost’?  Big thanks to the team at RNZ NineToNoon!

Should we give up copyright to beat China in the race for AI?

In conversation with Radio New Zealand’s Nine To Noon host Kathryn Ryan: Christie’s held its first auction of AI-generated art, earning a million dollars. Those AI artworks had been ‘trained’ from countless images, owned by other people. Is that legal? OpenAI and Google claim that unless they have free right to use – well, basically everything everywhere ever created by humanity – to train their AI models, the Chinese will win the AI race. Meanwhile, Hollywood’s A-listers called for protection of artists and their works against what they see as copyright theft. Plus: A Clockwork Orange comes to life for prisoners in solitary confinement – and is your chatbot flattering you?

Series 2024 – Halloween Special: Ghosts in the Machine

Google’s new NotebookLM digests documents to create AI summaries, and provides an incredible new way to search a vast set of information. One little throwaway feature became a viral hit – “Audio Overview”, which creates a ‘podcast’ featuring two clever, chipper – and entirely synthetic – personas, engaging in a ‘deep dive’ on the content within NotebookLM. In the spirit of Halloween, we uploaded the scripts for the first four episodes of Series 2024 – and NotebookLM rose to the occasion. Is this the future of podcasting? And — HAVE I JUST BEEN PUT OUT OF A JOB?

AI Miniseries SPECIAL EPISODE “Superpowers School Podcast with host Paddy Dhanda”

I really enjoy being a guest on other folks’ podcasts – all of the fun, almost none of the work. To help publicise my new book Getting Started with ChatGTP and AI Chatbots, I hopped onto Paddy Dhanda’s Superpowers School podcast for a 45 minute chat about AI chatbots, using AI safely & wisely – and what’s coming. It’s great fun, so have a listen!

If you’d like a signed copy of my book, just follow this link!

AI Miniseries #2 – “The Gunpowder Plot – How Meta put an AI chatbot on your smartphone and changed the game”

While everyone was going gaga over ChatGPT, Meta – the former Facebook – rewrote the AI playbook with LLaMA. Small enough to run on a smartphone, LLaMA gives us a glimpse of what the world will look like at the end of next year – with AI chatbots everywhere, inside almost everything. LLaMA and its tiny kin are amazing – but are they safe?